Software Quality Measurement: Assessment, Prediction and Validation

TR Number




Journal Title

Journal ISSN

Volume Title


Department of Computer Science, Virginia Polytechnic Institute & State University


Research addressing three challenging problems is described: (1) software quality assessment, (2) software quality prediction, and (3) the validation of both capabilities. The three are the focus of an empirical investigation conducted by a university research team using a development project by a Navy contractor as a laboratory. The predominantly Ada implementation total over 100,000 lines of source code, with approximately 40,000 included in this study. Using a procedure that couples two fundamental concepts: the Objectives/Principles/Attributes framework with software quality indicators, this longitudinal research project represents the most comprehensive effort yet to validate an approach to assessing software quality. Additionally, the more challenging problem of predicting software quality is tackled through an extraction and analysis of both process and product characteristics. Automated tools applied to both programs and documentation are described in the context of their contributing roles in measurement. Sources of difficulty in automating the assessment are identified. The basis for statistical validation are described, and anticipated hypotheses are examined.